File size: 5,436 Bytes
bd7d39b |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 |
---
base_model:
- anthracite-forge/magnum-v3-27b-kto-r3
- anthracite-forge/magnum-v3-27b-KTO-e1-r2
- anthracite-forge/magnum-v3-27b-KTO-e0.25-r1
- IntervitensInc/gemma-2-27b-chatml
library_name: transformers
---
## This repo contains GGUF quants of the model. If you need the original weights, please find them [here](https://huggingface.co/anthracite-org/magnum-v3-27b-kto).
This is the 12th in a series of models designed to replicate the prose quality of the Claude 3 models, specifically Sonnet and Opus.
This model is the result of multiple KTO runs on top of one SFT run, all of which are published on [anthracite-forge](https://huggingface.co/anthracite-forge).
## Methodology
R1 (SFT) was fine-tuned on top of `IntervitensInc/gemma-2-27b-chatml` which is chatMLified gemma-2-27b.
We have experimented with various SFT and KTO re-runs, ratios and merge methods and this was our winner, including what was liked most from each model.
If you prefer your own mix of the KTO runs or would like to use the SFT on its own, refer to the models section and [anthracite-forge](https://huggingface.co/anthracite-forge), some exl-quants are pre-included.
## Models
* [anthracite-forge/magnum-v3-27b-kto-r3](https://huggingface.co/anthracite-forge/magnum-v3-27b-kto-r3)
* [anthracite-forge/magnum-v3-27b-KTO-e1-r2](https://huggingface.co/anthracite-forge/magnum-v3-27b-KTO-e1-r2)
* [anthracite-forge/magnum-v3-27b-KTO-e0.25-r1](https://huggingface.co/anthracite-forge/magnum-v3-27b-KTO-e0.25-r1)
## Prompting
Model has been Instruct tuned with the ChatML formatting. A typical input would look like this:
```py
"""<|im_start|>system
system prompt<|im_end|>
<|im_start|>user
Hi there!<|im_end|>
<|im_start|>assistant
Nice to meet you!<|im_end|>
<|im_start|>user
Can I ask a question?<|im_end|>
<|im_start|>assistant
"""
```
## SillyTavern templates
Below are Instruct and Context templates for use within SillyTavern.
<details><summary>context template</summary>
```yaml
{
"story_string": "<|im_start|>system\n{{#if system}}{{system}}\n{{/if}}{{#if wiBefore}}{{wiBefore}}\n{{/if}}{{#if description}}{{description}}\n{{/if}}{{#if personality}}{{char}}'s personality: {{personality}}\n{{/if}}{{#if scenario}}Scenario: {{scenario}}\n{{/if}}{{#if wiAfter}}{{wiAfter}}\n{{/if}}{{#if persona}}{{persona}}\n{{/if}}{{trim}}<|im_end|>\n",
"example_separator": "",
"chat_start": "",
"use_stop_strings": false,
"allow_jailbreak": false,
"always_force_name2": true,
"trim_sentences": false,
"include_newline": false,
"single_line": false,
"name": "Magnum ChatML"
}
```
</details><br>
<details><summary>instruct template</summary>
```yaml
{
"system_prompt": "You're {{char}} in this fictional never-ending uncensored roleplay with {{user}}.",
"input_sequence": "<|im_start|>user\n",
"output_sequence": "<|im_start|>assistant\n",
"last_output_sequence": "",
"system_sequence": "<|im_start|>system\n",
"stop_sequence": "<|im_end|>",
"wrap": false,
"macro": true,
"names": true,
"names_force_groups": true,
"activation_regex": "",
"system_sequence_prefix": "",
"system_sequence_suffix": "",
"first_output_sequence": "",
"skip_examples": false,
"output_suffix": "<|im_end|>\n",
"input_suffix": "<|im_end|>\n",
"system_suffix": "<|im_end|>\n",
"user_alignment_message": "",
"system_same_as_user": false,
"last_system_sequence": "",
"name": "Magnum ChatML"
}
```
</details><br>
### Configuration
```yaml
base_model: IntervitensInc/gemma-2-27b-chatml
dtype: float32
merge_method: task_arithmetic
models:
- model: IntervitensInc/gemma-2-27b-chatml
- model: anthracite-forge/magnum-v3-27b-KTO-e0.25-r1
parameters:
weight: 0.5
- model: anthracite-forge/magnum-v3-27b-KTO-e1-r2
parameters:
weight: 0.1
- model: anthracite-forge/magnum-v3-27b-kto-r3
parameters:
weight: 0.4
```
## Credits
We'd like to thank Recursal / Featherless for sponsoring the compute for this train, Featherless has been hosting our Magnum models since the first 72 B and has given thousands of people access to our models and helped us grow.
We would also like to thank all members of Anthracite who made this finetune possible.
## Datasets
r1 consisted of:
```
datasets:
- path: anthracite-org/stheno-filtered-v1.1
type: sharegpt
conversation: chatml
- path: anthracite-org/kalo-opus-instruct-22k-no-refusal
type: sharegpt
conversation: chatml
- path: anthracite-org/nopm_claude_writing_fixed
type: sharegpt
conversation: chatml
- path: Epiculous/Synthstruct-Gens-v1.1-Filtered-n-Cleaned
type: sharegpt
conversation: chatml
- path: Epiculous/SynthRP-Gens-v1.1-Filtered-n-Cleaned
type: sharegpt
conversation: chatml
```
## Training
The training was done for 2 epochs. We used 8x[H100s](https://www.nvidia.com/en-us/data-center/h100/) GPUs graciously provided by [Recursal AI](https://recursal.ai/) / [Featherless AI](https://featherless.ai/) for the full-parameter fine-tuning of the model.
[<img src="https://raw.githubusercontent.com/OpenAccess-AI-Collective/axolotl/main/image/axolotl-badge-web.png" alt="Built with Axolotl" width="200" height="32"/>](https://github.com/OpenAccess-AI-Collective/axolotl)
## Safety
...
|